Towards planbased automatic exploration of virtual worlds

Abstract

In order to efficiently explore unknown virtual worlds with a camera, the camera has to be as autonomous as possible, in order to take decisions whenever it is necessary. In this paper, we present a formalisation of the automatic virtual world exploration problem, together with a technique allowing online incremental external intelligent exploration of a virtual world by a camera. This technique is based on an evaluation function which takes into account scene's covering and quality of points of view, together with a small number of positions that should be reached by the camera and tries to make exploration plans in order to reach them using incremental determination of the next camera position.